Segmentación de imagenes: Método de conjunto de nivel

Análisis Numérico de Ecuaciones en Derivadas Parciales:Teoría y Laboratorio Autores: Felipe Urrutia, Daniel Minaya, Sebastian Toloza Tutor: Axel Osses A., Auxiliar: Emir N. Chacra

Resultados avanzados

Estudio del conjunto de datos PASCAL 2012 para la segmentacion de imagenes

import platform
from class_levelsetmethod import LevelSetMethod, Image, np, plt, animation, os, ImageOps, ListedColormap, pickle
platform.processor()

'AMD64 Family 23 Model 113 Stepping 0, AuthenticAMD'

def get_groud_truth(id_name):
    img_r = Image.open(f"dataset/database_pascal/{id_name}.png").convert("L")

    dim_r = (img_r.size[0], img_r.size[1])
    xy_dim_max = np.argmax(dim_r)
    if dim_r[xy_dim_max] > 100:
        basewidth = 100
        w, h = np.array([int(dim_r[_]*100/dim_r[xy_dim_max]) for _ in range(2)])
        img_r = img_r.resize((w,h), Image.ANTIALIAS)


    dim_r = (img_r.size[1], img_r.size[0])

    img_r = ImageOps.posterize(img_r, 4)
    img_r = np.array(img_r.getdata()).reshape((dim_r[0], dim_r[1]))
    img_r = (255*(img_r-np.min(img_r))/(np.max(img_r)-np.min(img_r))).astype(int)

    img_r = ((img_r>np.mean(img_r))).astype(int)
    return img_r

def precision_recall(y_real, y_pred, pos_label=1):
    tp = np.sum( (y_real==y_pred) & (y_pred==pos_label) )
    p = np.sum( y_real==pos_label )
    recall = tp/p
    
    fp = np.sum( (y_real!=y_pred) & (y_pred==pos_label) )
    precision = tp/(tp+fp)
    
    return precision, recall

def plot_precision_recall(img_r, iterations):
    clf = [(un>0).astype(int) for un in iterations]
    
    precision_recall_curve = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf]
    precision = np.array([prc_k[0] for prc_k in precision_recall_curve])
    recall = np.array([prc_k[1] for prc_k in precision_recall_curve])
    
    mask_0 = np.isclose(precision+recall, 0)
    mask_1 = ~mask_0
    
    best_iter= np.argmax( mask_1*(2*precision*recall)/(precision+recall+mask_0) )
    
    fig, ax = plt.subplots(1, 1, figsize=(5, 5))
    
    ax.scatter(precision, recall, c=range(len(precision_recall_curve)), cmap="plasma")
    ax.scatter([precision[best_iter]], [recall[best_iter]], c="red", s=50, label=f"Iter: {best_iter}\nRecall: {100*recall[best_iter]: .1f}\%\nPrecision: {100*precision[best_iter]: .1f}\%")
    ax.set_yticks(np.linspace(3/4, 1, 5).round(2));
    ax.set_xticks(np.linspace(3/4, 1, 5).round(2));
    
    plt_eps = 0.01
    plt.xticks(fontsize=18*0.8)
    plt.yticks(fontsize=18*0.8)
    
    ax.set_xlim([3/4-plt_eps, 1+plt_eps])
    ax.set_ylim([3/4-plt_eps, 1+plt_eps])
    ax.set_ylabel("Recall", fontsize=18)
    ax.set_xlabel("Precision", fontsize=18)
    
    plt.grid()
    ax.legend(fontsize=18*0.8)
    
    ax.set_title("Precision vs Recall", fontsize=18*1.5);
    
    plt.gca().set_aspect('equal')

%%time
pascal_images = ["2009_001314", "2008_004621", "2008_002247", "2007_004052", "2009_005078"]
# path_save_file_per_test = {}
dic_img_arr = {}
for id_name in pascal_images:
    img_o = Image.open(f"dataset/database_pascal/{id_name}.jpg").convert("L")
    dim = (img_o.size[1], img_o.size[0])
    img = np.array(img_o.getdata()).reshape((dim[0], dim[1]))
    dic_img_arr[id_name] = img

Wall time: 77.6 ms

%%time
id_name = pascal_images[0]
segmentator = LevelSetMethod(id_name=id_name)
segmentator.set_image(
    dic_img_arr[id_name], 
    reshape=True,
    norm=True
)
segmentator.show_image(grid=True)

<set_image> img: <class 'numpy.ndarray'>, reshape: True, norm: True
|img_dim:     [457 500]
|img_min/max: (16, 195)
|new_img_min/max(resize): (1.0477378966863588e-08, 5.991934994504968e-08)
|new_dim:     [ 91 100]
|new_img_min/max(norm): (0, 255)
|size_domain: 9100
</set_image>
Wall time: 17 ms

%%time
segmentator.eval_stopfunct(
    list_sgm=np.array([0.01, 0.5, 1]), 
    list_lmbd=np.array([15, 5, 1]),
    cmap="viridis_r"
)
plt.tight_layout()

Wall time: 166 ms

%%time
segmentator.set_stopfunct(
    sgm=0.01, 
    lmbd=5,
)

<set_stopfunct> sigma: 0.01, lambda: 5
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (0.0006260956674179814, 1.0)
</set_stopfunct>
Wall time: 2 ms

%%time
segmentator.eval_sgndist(
    c=np.array([0.4, 0.5]), 
    r=0.18,
    e=1
)

Wall time: 18 ms

%%time
segmentator.set_sgndist(
    c=np.array([0.4, 0.5]), 
    r=0.18,
    e=1
)

<set_sgndist> centro: [0.4 0.5], radio: 0.18, excentricidad: 1
</set_sgndist>
Wall time: 1 ms

%%time 
segmentator.show_setting(cmap_stopfunct="viridis_r")

Wall time: 41 ms

%%time 
segmentator.print_setting()

<print_setting>
|sigma: 0.01
|lambda: 5
|centro: [0.4 0.5]
|radio: 0.18
|excentricidad: 1
</print_setting>
Wall time: 1e+03 µs

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2009_001314__iterations_times__sgm_0_01__lmbd_5__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
|Iter: 1, T iter: 2.2s, T trans:  4.1s, T per iter: 2.0s p/m  0.2s, T rest:  307.4s p/m  2.1s
|Iter: 6, T iter: 1.9s, T trans:  13.5s, T per iter: 1.9s p/m  0.1s, T rest:  280.1s p/m  1.4s
|Iter: 11, T iter: 2.3s, T trans:  23.4s, T per iter: 1.9s p/m  0.1s, T rest:  272.8s p/m  1.6s
|Iter: 16, T iter: 1.9s, T trans:  32.9s, T per iter: 1.9s p/m  0.1s, T rest:  261.1s p/m  1.3s
|Iter: 21, T iter: 2.2s, T trans:  42.7s, T per iter: 1.9s p/m  0.1s, T rest:  252.5s p/m  1.4s
|Iter: 26, T iter: 1.9s, T trans:  52.3s, T per iter: 1.9s p/m  0.1s, T rest:  242.0s p/m  1.2s
|Iter: 31, T iter: 1.9s, T trans:  61.8s, T per iter: 1.9s p/m  0.1s, T rest:  231.6s p/m  1.1s
|Iter: 36, T iter: 1.9s, T trans:  71.3s, T per iter: 1.9s p/m  0.1s, T rest:  221.5s p/m  1.0s
|Iter: 41, T iter: 2.3s, T trans:  81.1s, T per iter: 1.9s p/m  0.1s, T rest:  212.5s p/m  1.1s
|Iter: 46, T iter: 2.0s, T trans:  90.8s, T per iter: 1.9s p/m  0.1s, T rest:  202.8s p/m  1.1s
|Iter: 51, T iter: 2.0s, T trans:  100.5s, T per iter: 1.9s p/m  0.1s, T rest:  193.3s p/m  1.0s
|Iter: 56, T iter: 2.0s, T trans:  110.5s, T per iter: 1.9s p/m  0.1s, T rest:  184.1s p/m  0.9s
|Iter: 61, T iter: 2.0s, T trans:  120.6s, T per iter: 1.9s p/m  0.1s, T rest:  175.1s p/m  0.9s
|Iter: 66, T iter: 2.0s, T trans:  130.6s, T per iter: 1.9s p/m  0.1s, T rest:  165.7s p/m  0.9s
|Iter: 71, T iter: 2.4s, T trans:  140.7s, T per iter: 2.0s p/m  0.1s, T rest:  156.4s p/m  0.9s
|Iter: 76, T iter: 2.0s, T trans:  150.5s, T per iter: 2.0s p/m  0.1s, T rest:  146.6s p/m  0.9s
|Iter: 81, T iter: 2.3s, T trans:  160.6s, T per iter: 2.0s p/m  0.1s, T rest:  137.1s p/m  0.9s
|Iter: 86, T iter: 1.9s, T trans:  170.2s, T per iter: 2.0s p/m  0.1s, T rest:  127.2s p/m  0.8s
|Iter: 91, T iter: 2.3s, T trans:  180.3s, T per iter: 2.0s p/m  0.1s, T rest:  117.6s p/m  0.8s
|Iter: 96, T iter: 1.9s, T trans:  190.1s, T per iter: 2.0s p/m  0.1s, T rest:  107.8s p/m  0.8s
|Iter: 101, T iter: 2.3s, T trans:  200.2s, T per iter: 2.0s p/m  0.1s, T rest:  98.1s p/m  0.7s
|Iter: 106, T iter: 1.9s, T trans:  209.9s, T per iter: 2.0s p/m  0.1s, T rest:  88.3s p/m  0.7s
|Iter: 111, T iter: 2.0s, T trans:  219.6s, T per iter: 2.0s p/m  0.1s, T rest:  78.4s p/m  0.6s
|Iter: 116, T iter: 1.9s, T trans:  229.3s, T per iter: 2.0s p/m  0.1s, T rest:  68.6s p/m  0.6s
|Iter: 121, T iter: 2.0s, T trans:  239.1s, T per iter: 2.0s p/m  0.1s, T rest:  58.8s p/m  0.5s
|Iter: 126, T iter: 1.9s, T trans:  248.8s, T per iter: 2.0s p/m  0.1s, T rest:  49.0s p/m  0.5s
|Iter: 131, T iter: 2.0s, T trans:  258.8s, T per iter: 2.0s p/m  0.1s, T rest:  39.2s p/m  0.4s
|Iter: 136, T iter: 1.9s, T trans:  268.5s, T per iter: 2.0s p/m  0.1s, T rest:  29.4s p/m  0.4s
|Iter: 141, T iter: 2.5s, T trans:  278.9s, T per iter: 2.0s p/m  0.1s, T rest:  19.6s p/m  0.3s
|Iter: 146, T iter: 1.9s, T trans:  288.6s, T per iter: 2.0s p/m  0.1s, T rest:  9.8s p/m  0.2s
</solve>
Wall time: 4min 56s

%%time
segmentator.show_solution(
    show_iters=[26, 49, 99]
)

Wall time: 69.1 ms

path_save_file_per_test["test_0"] = segmentator.path_save_file

img_r = get_groud_truth(segmentator.id_name)
plt.imshow(img_r, cmap=ListedColormap(['black','tab:red']));

plot_precision_recall(img_r, segmentator.iterations)

segmentator.show_solution(
    show_iters=[45, 55, 65]
)
plt.tight_layout()

%%time
segmentator.set_stopfunct(
    sgm=0.01, 
    lmbd=15,
)

<set_stopfunct> sigma: 0.01, lambda: 15
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (0.005606777971592324, 1.0)
</set_stopfunct>
Wall time: 1.54 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2009_001314__iterations_times__sgm_0_01__lmbd_15__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
|Iter: 1, T iter: 2.0s, T trans:  4.0s, T per iter: 2.0s p/m  0.0s, T rest:  296.8s p/m  0.0s
|Iter: 6, T iter: 2.0s, T trans:  13.8s, T per iter: 2.0s p/m  0.0s, T rest:  286.4s p/m  0.1s
|Iter: 11, T iter: 2.0s, T trans:  23.6s, T per iter: 2.0s p/m  0.0s, T rest:  275.6s p/m  0.1s
|Iter: 16, T iter: 1.9s, T trans:  33.5s, T per iter: 2.0s p/m  0.0s, T rest:  265.9s p/m  0.2s
|Iter: 21, T iter: 2.0s, T trans:  43.4s, T per iter: 2.0s p/m  0.0s, T rest:  256.5s p/m  0.3s
|Iter: 26, T iter: 1.9s, T trans:  53.1s, T per iter: 2.0s p/m  0.0s, T rest:  245.8s p/m  0.3s
|Iter: 31, T iter: 2.0s, T trans:  62.9s, T per iter: 2.0s p/m  0.0s, T rest:  235.9s p/m  0.3s
|Iter: 36, T iter: 1.9s, T trans:  72.7s, T per iter: 2.0s p/m  0.0s, T rest:  225.9s p/m  0.3s
|Iter: 41, T iter: 2.0s, T trans:  82.4s, T per iter: 2.0s p/m  0.0s, T rest:  215.8s p/m  0.3s
|Iter: 46, T iter: 1.9s, T trans:  92.1s, T per iter: 2.0s p/m  0.0s, T rest:  205.7s p/m  0.3s
|Iter: 51, T iter: 2.3s, T trans:  102.1s, T per iter: 2.0s p/m  0.1s, T rest:  196.4s p/m  0.6s
|Iter: 56, T iter: 2.0s, T trans:  112.0s, T per iter: 2.0s p/m  0.1s, T rest:  186.6s p/m  0.5s
|Iter: 61, T iter: 2.3s, T trans:  122.1s, T per iter: 2.0s p/m  0.1s, T rest:  177.2s p/m  0.6s
|Iter: 66, T iter: 1.9s, T trans:  131.7s, T per iter: 2.0s p/m  0.1s, T rest:  167.1s p/m  0.6s
|Iter: 71, T iter: 2.3s, T trans:  141.7s, T per iter: 2.0s p/m  0.1s, T rest:  157.5s p/m  0.7s
|Iter: 76, T iter: 1.9s, T trans:  151.4s, T per iter: 2.0s p/m  0.1s, T rest:  147.4s p/m  0.6s
|Iter: 81, T iter: 2.0s, T trans:  161.1s, T per iter: 2.0s p/m  0.1s, T rest:  137.5s p/m  0.6s
|Iter: 86, T iter: 2.0s, T trans:  171.0s, T per iter: 2.0s p/m  0.1s, T rest:  127.8s p/m  0.6s
|Iter: 91, T iter: 2.0s, T trans:  181.0s, T per iter: 2.0s p/m  0.1s, T rest:  118.0s p/m  0.5s
|Iter: 96, T iter: 1.9s, T trans:  190.7s, T per iter: 2.0s p/m  0.1s, T rest:  108.1s p/m  0.5s
|Iter: 101, T iter: 2.4s, T trans:  200.8s, T per iter: 2.0s p/m  0.1s, T rest:  98.4s p/m  0.6s
|Iter: 106, T iter: 2.0s, T trans:  210.5s, T per iter: 2.0s p/m  0.1s, T rest:  88.5s p/m  0.5s
|Iter: 111, T iter: 2.0s, T trans:  220.3s, T per iter: 2.0s p/m  0.1s, T rest:  78.7s p/m  0.5s
|Iter: 116, T iter: 1.9s, T trans:  230.0s, T per iter: 2.0s p/m  0.1s, T rest:  68.8s p/m  0.4s
|Iter: 121, T iter: 2.4s, T trans:  240.1s, T per iter: 2.0s p/m  0.1s, T rest:  59.0s p/m  0.4s
|Iter: 126, T iter: 1.9s, T trans:  249.8s, T per iter: 2.0s p/m  0.1s, T rest:  49.2s p/m  0.4s
|Iter: 131, T iter: 2.4s, T trans:  260.1s, T per iter: 2.0s p/m  0.1s, T rest:  39.4s p/m  0.4s
|Iter: 136, T iter: 1.9s, T trans:  269.9s, T per iter: 2.0s p/m  0.1s, T rest:  29.6s p/m  0.3s
|Iter: 141, T iter: 2.0s, T trans:  279.8s, T per iter: 2.0s p/m  0.1s, T rest:  19.7s p/m  0.3s
|Iter: 146, T iter: 1.9s, T trans:  289.4s, T per iter: 2.0s p/m  0.1s, T rest:  9.8s p/m  0.2s
</solve>
Wall time: 4min 57s

%%time
segmentator.show_solution(
    show_iters=[45, 55, 65]
)
plt.tight_layout()

Wall time: 82.1 ms

path_save_file_per_test["test_1"] = segmentator.path_save_file

plot_precision_recall(img_r, segmentator.iterations)

%%time
segmentator.set_stopfunct(
    sgm=0.01, 
    lmbd=1,
)

<set_stopfunct> sigma: 0.01, lambda: 1
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (2.5058888387711116e-05, 1.0)
</set_stopfunct>
Wall time: 999 µs

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2009_001314__iterations_times__sgm_0_01__lmbd_1__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
|Iter: 1, T iter: 2.0s, T trans:  4.0s, T per iter: 2.0s p/m  0.0s, T rest:  300.3s p/m  0.5s
|Iter: 6, T iter: 2.0s, T trans:  14.2s, T per iter: 2.0s p/m  0.1s, T rest:  294.2s p/m  0.7s
|Iter: 11, T iter: 2.0s, T trans:  24.2s, T per iter: 2.0s p/m  0.0s, T rest:  282.0s p/m  0.5s
|Iter: 16, T iter: 2.1s, T trans:  34.5s, T per iter: 2.0s p/m  0.1s, T rest:  274.1s p/m  0.6s
|Iter: 21, T iter: 2.3s, T trans:  44.6s, T per iter: 2.0s p/m  0.1s, T rest:  263.6s p/m  0.9s
|Iter: 26, T iter: 2.0s, T trans:  54.4s, T per iter: 2.0s p/m  0.1s, T rest:  251.8s p/m  0.9s
|Iter: 31, T iter: 2.4s, T trans:  64.8s, T per iter: 2.0s p/m  0.1s, T rest:  242.9s p/m  1.1s
|Iter: 36, T iter: 2.0s, T trans:  74.8s, T per iter: 2.0s p/m  0.1s, T rest:  232.4s p/m  1.0s
|Iter: 41, T iter: 2.3s, T trans:  84.9s, T per iter: 2.0s p/m  0.1s, T rest:  222.3s p/m  1.1s
|Iter: 46, T iter: 1.9s, T trans:  94.6s, T per iter: 2.0s p/m  0.1s, T rest:  211.4s p/m  1.0s
|Iter: 51, T iter: 2.0s, T trans:  104.3s, T per iter: 2.0s p/m  0.1s, T rest:  200.7s p/m  1.0s
|Iter: 56, T iter: 1.9s, T trans:  114.0s, T per iter: 2.0s p/m  0.1s, T rest:  190.0s p/m  0.9s
|Iter: 61, T iter: 2.0s, T trans:  123.7s, T per iter: 2.0s p/m  0.1s, T rest:  179.6s p/m  0.9s
|Iter: 66, T iter: 1.9s, T trans:  133.4s, T per iter: 2.0s p/m  0.1s, T rest:  169.2s p/m  0.9s
|Iter: 71, T iter: 2.0s, T trans:  143.1s, T per iter: 2.0s p/m  0.1s, T rest:  159.0s p/m  0.8s
|Iter: 76, T iter: 1.9s, T trans:  152.8s, T per iter: 2.0s p/m  0.1s, T rest:  148.8s p/m  0.8s
|Iter: 81, T iter: 2.0s, T trans:  162.5s, T per iter: 2.0s p/m  0.1s, T rest:  138.7s p/m  0.7s
|Iter: 86, T iter: 1.9s, T trans:  172.1s, T per iter: 2.0s p/m  0.1s, T rest:  128.6s p/m  0.7s
|Iter: 91, T iter: 2.0s, T trans:  181.8s, T per iter: 2.0s p/m  0.1s, T rest:  118.6s p/m  0.6s
|Iter: 96, T iter: 1.9s, T trans:  191.5s, T per iter: 2.0s p/m  0.1s, T rest:  108.6s p/m  0.6s
|Iter: 101, T iter: 2.0s, T trans:  201.3s, T per iter: 2.0s p/m  0.1s, T rest:  98.7s p/m  0.6s
|Iter: 106, T iter: 1.9s, T trans:  210.9s, T per iter: 2.0s p/m  0.1s, T rest:  88.7s p/m  0.5s
|Iter: 111, T iter: 2.0s, T trans:  220.6s, T per iter: 2.0s p/m  0.1s, T rest:  78.8s p/m  0.5s
|Iter: 116, T iter: 1.9s, T trans:  230.2s, T per iter: 2.0s p/m  0.1s, T rest:  68.9s p/m  0.5s
|Iter: 121, T iter: 2.0s, T trans:  239.9s, T per iter: 2.0s p/m  0.1s, T rest:  59.0s p/m  0.4s
|Iter: 126, T iter: 1.9s, T trans:  249.6s, T per iter: 2.0s p/m  0.1s, T rest:  49.1s p/m  0.4s
|Iter: 131, T iter: 2.4s, T trans:  259.7s, T per iter: 2.0s p/m  0.1s, T rest:  39.3s p/m  0.4s
|Iter: 136, T iter: 1.9s, T trans:  269.4s, T per iter: 2.0s p/m  0.1s, T rest:  29.5s p/m  0.3s
|Iter: 141, T iter: 2.4s, T trans:  279.4s, T per iter: 2.0s p/m  0.1s, T rest:  19.7s p/m  0.3s
|Iter: 146, T iter: 1.9s, T trans:  289.1s, T per iter: 2.0s p/m  0.1s, T rest:  9.8s p/m  0.2s
</solve>
Wall time: 4min 57s

%%time
segmentator.show_solution(
    show_iters=[45, 55, 65]
)
plt.tight_layout()

Wall time: 83.1 ms

path_save_file_per_test["test_2"] = segmentator.path_save_file

plot_precision_recall(img_r, segmentator.iterations)

%%time
segmentator.set_stopfunct(
    sgm=1, 
    lmbd=5,
)

<set_stopfunct> sigma: 1, lambda: 5
|gauss_min/max: (11, 213)
|stopfunct_min/max:     (0.005533421868083223, 1.0)
</set_stopfunct>
Wall time: 2 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2009_001314__iterations_times__sgm_1__lmbd_5__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
|Iter: 1, T iter: 1.9s, T trans:  3.8s, T per iter: 1.9s p/m  0.0s, T rest:  285.9s p/m  0.4s
|Iter: 6, T iter: 1.9s, T trans:  13.5s, T per iter: 1.9s p/m  0.0s, T rest:  279.0s p/m  0.3s
|Iter: 11, T iter: 1.9s, T trans:  23.2s, T per iter: 1.9s p/m  0.0s, T rest:  270.3s p/m  0.3s
|Iter: 16, T iter: 1.9s, T trans:  32.9s, T per iter: 1.9s p/m  0.0s, T rest:  260.9s p/m  0.2s
|Iter: 21, T iter: 2.0s, T trans:  42.6s, T per iter: 1.9s p/m  0.0s, T rest:  251.8s p/m  0.3s
|Iter: 26, T iter: 2.0s, T trans:  52.4s, T per iter: 1.9s p/m  0.0s, T rest:  242.5s p/m  0.2s
|Iter: 31, T iter: 1.9s, T trans:  62.0s, T per iter: 1.9s p/m  0.0s, T rest:  232.6s p/m  0.2s
|Iter: 36, T iter: 2.0s, T trans:  71.8s, T per iter: 1.9s p/m  0.0s, T rest:  223.1s p/m  0.2s
|Iter: 41, T iter: 2.4s, T trans:  82.1s, T per iter: 2.0s p/m  0.1s, T rest:  214.9s p/m  0.8s
|Iter: 46, T iter: 1.9s, T trans:  91.8s, T per iter: 2.0s p/m  0.1s, T rest:  205.0s p/m  0.7s
|Iter: 51, T iter: 2.2s, T trans:  101.8s, T per iter: 2.0s p/m  0.1s, T rest:  195.7s p/m  0.8s
|Iter: 56, T iter: 1.9s, T trans:  111.4s, T per iter: 2.0s p/m  0.1s, T rest:  185.7s p/m  0.8s
|Iter: 61, T iter: 2.3s, T trans:  121.4s, T per iter: 2.0s p/m  0.1s, T rest:  176.2s p/m  0.8s
|Iter: 66, T iter: 2.0s, T trans:  131.2s, T per iter: 2.0s p/m  0.1s, T rest:  166.4s p/m  0.8s
|Iter: 71, T iter: 2.3s, T trans:  141.2s, T per iter: 2.0s p/m  0.1s, T rest:  156.8s p/m  0.8s
|Iter: 76, T iter: 1.9s, T trans:  150.8s, T per iter: 2.0s p/m  0.1s, T rest:  146.9s p/m  0.7s
|Iter: 81, T iter: 2.2s, T trans:  160.8s, T per iter: 2.0s p/m  0.1s, T rest:  137.3s p/m  0.7s
|Iter: 86, T iter: 1.9s, T trans:  170.6s, T per iter: 2.0s p/m  0.1s, T rest:  127.4s p/m  0.7s
|Iter: 91, T iter: 2.3s, T trans:  180.7s, T per iter: 2.0s p/m  0.1s, T rest:  117.8s p/m  0.7s
|Iter: 96, T iter: 1.9s, T trans:  190.4s, T per iter: 2.0s p/m  0.1s, T rest:  108.0s p/m  0.7s
|Iter: 101, T iter: 2.3s, T trans:  200.5s, T per iter: 2.0s p/m  0.1s, T rest:  98.3s p/m  0.7s
|Iter: 106, T iter: 2.0s, T trans:  210.3s, T per iter: 2.0s p/m  0.1s, T rest:  88.4s p/m  0.6s
|Iter: 111, T iter: 2.0s, T trans:  220.0s, T per iter: 2.0s p/m  0.1s, T rest:  78.6s p/m  0.6s
|Iter: 116, T iter: 2.0s, T trans:  229.8s, T per iter: 2.0s p/m  0.1s, T rest:  68.7s p/m  0.5s
|Iter: 121, T iter: 2.1s, T trans:  239.7s, T per iter: 2.0s p/m  0.1s, T rest:  58.9s p/m  0.5s
|Iter: 126, T iter: 1.9s, T trans:  249.4s, T per iter: 2.0s p/m  0.1s, T rest:  49.1s p/m  0.4s
|Iter: 131, T iter: 2.1s, T trans:  259.2s, T per iter: 2.0s p/m  0.1s, T rest:  39.3s p/m  0.4s
|Iter: 136, T iter: 2.0s, T trans:  269.0s, T per iter: 2.0s p/m  0.1s, T rest:  29.4s p/m  0.3s
|Iter: 141, T iter: 2.1s, T trans:  278.9s, T per iter: 2.0s p/m  0.1s, T rest:  19.6s p/m  0.3s
|Iter: 146, T iter: 2.0s, T trans:  288.7s, T per iter: 2.0s p/m  0.1s, T rest:  9.8s p/m  0.2s
</solve>
Wall time: 4min 56s

%%time
segmentator.show_solution(
    show_iters=[45, 55, 65]
)
plt.tight_layout()

Wall time: 83.1 ms

path_save_file_per_test["test_3"] = segmentator.path_save_file

plot_precision_recall(img_r, segmentator.iterations)

%%time
id_name = pascal_images[1]
segmentator = LevelSetMethod(id_name=id_name)
segmentator.set_image(
    dic_img_arr[id_name], 
    reshape=True,
    norm=True
)
segmentator.show_image(grid=True)

<set_image> img: <class 'numpy.ndarray'>, reshape: True, norm: True
|img_dim:     [375 500]
|img_min/max: (0, 147)
|new_img_min/max(resize): (2.5611370807889267e-09, 6.44940883071381e-08)
|new_dim:     [ 75 100]
|new_img_min/max(norm): (0, 255)
|size_domain: 7500
</set_image>
Wall time: 20 ms

%%time
segmentator.eval_stopfunct(
    list_sgm=np.array([0.01, 0.5, 1]), 
    list_lmbd=np.array([15, 10, 5]),
    cmap="viridis_r"
)
plt.tight_layout()

Wall time: 704 ms

%%time
segmentator.set_stopfunct(
    sgm=0.5, 
    lmbd=5,
)

<set_stopfunct> sigma: 0.5, lambda: 5
|gauss_min/max: (3, 251)
|stopfunct_min/max:     (0.0014478484971332604, 1.0)
</set_stopfunct>
Wall time: 2 ms

%%time
segmentator.eval_sgndist(
    c=np.array([0.5, 0.53]), 
    r=0.18,
    e=1
)

Wall time: 21 ms

%%time
segmentator.set_sgndist(
    c=np.array([0.5, 0.53]), 
    r=0.18,
    e=1
)

<set_sgndist> centro: [0.5  0.53], radio: 0.18, excentricidad: 1
</set_sgndist>
Wall time: 2 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2008_004621__iterations_times__sgm_0_5__lmbd_5__cx_0_5__cy_0_53__r_0_18__e_1__tau_0_1__k_-5
|Iter: 1, T iter: 1.9s, T trans:  3.5s, T per iter: 1.7s p/m  0.2s, T rest:  260.0s p/m  2.1s
|Iter: 6, T iter: 1.7s, T trans:  11.4s, T per iter: 1.6s p/m  0.1s, T rest:  237.0s p/m  1.4s
|Iter: 11, T iter: 1.6s, T trans:  19.5s, T per iter: 1.6s p/m  0.1s, T rest:  228.0s p/m  1.1s
|Iter: 16, T iter: 1.6s, T trans:  27.4s, T per iter: 1.6s p/m  0.1s, T rest:  217.5s p/m  1.0s
|Iter: 21, T iter: 1.6s, T trans:  35.3s, T per iter: 1.6s p/m  0.1s, T rest:  208.7s p/m  0.9s
|Iter: 26, T iter: 1.6s, T trans:  43.1s, T per iter: 1.6s p/m  0.1s, T rest:  199.5s p/m  0.8s
|Iter: 31, T iter: 1.6s, T trans:  51.0s, T per iter: 1.6s p/m  0.1s, T rest:  191.1s p/m  0.7s
|Iter: 36, T iter: 1.6s, T trans:  58.9s, T per iter: 1.6s p/m  0.1s, T rest:  183.2s p/m  0.7s
|Iter: 41, T iter: 2.0s, T trans:  67.4s, T per iter: 1.6s p/m  0.1s, T rest:  176.6s p/m  0.9s
|Iter: 46, T iter: 1.6s, T trans:  75.3s, T per iter: 1.6s p/m  0.1s, T rest:  168.1s p/m  0.9s
|Iter: 51, T iter: 1.6s, T trans:  83.4s, T per iter: 1.6s p/m  0.1s, T rest:  160.4s p/m  0.8s
|Iter: 56, T iter: 1.6s, T trans:  91.4s, T per iter: 1.6s p/m  0.1s, T rest:  152.3s p/m  0.8s
|Iter: 61, T iter: 1.6s, T trans:  99.3s, T per iter: 1.6s p/m  0.1s, T rest:  144.2s p/m  0.7s
|Iter: 66, T iter: 1.6s, T trans:  107.2s, T per iter: 1.6s p/m  0.1s, T rest:  136.0s p/m  0.7s
|Iter: 71, T iter: 1.6s, T trans:  115.0s, T per iter: 1.6s p/m  0.1s, T rest:  127.8s p/m  0.7s
|Iter: 76, T iter: 1.6s, T trans:  122.8s, T per iter: 1.6s p/m  0.1s, T rest:  119.6s p/m  0.6s
|Iter: 81, T iter: 1.7s, T trans:  131.0s, T per iter: 1.6s p/m  0.1s, T rest:  111.8s p/m  0.6s
|Iter: 86, T iter: 1.6s, T trans:  139.0s, T per iter: 1.6s p/m  0.1s, T rest:  103.9s p/m  0.6s
|Iter: 91, T iter: 1.7s, T trans:  147.0s, T per iter: 1.6s p/m  0.1s, T rest:  95.9s p/m  0.5s
|Iter: 96, T iter: 1.6s, T trans:  155.0s, T per iter: 1.6s p/m  0.1s, T rest:  87.9s p/m  0.5s
|Iter: 101, T iter: 1.6s, T trans:  162.9s, T per iter: 1.6s p/m  0.1s, T rest:  79.8s p/m  0.5s
|Iter: 106, T iter: 1.6s, T trans:  170.8s, T per iter: 1.6s p/m  0.1s, T rest:  71.8s p/m  0.4s
|Iter: 111, T iter: 1.6s, T trans:  178.7s, T per iter: 1.6s p/m  0.1s, T rest:  63.8s p/m  0.4s
|Iter: 116, T iter: 1.6s, T trans:  186.5s, T per iter: 1.6s p/m  0.1s, T rest:  55.8s p/m  0.4s
|Iter: 121, T iter: 1.6s, T trans:  194.5s, T per iter: 1.6s p/m  0.1s, T rest:  47.8s p/m  0.3s
|Iter: 126, T iter: 1.6s, T trans:  202.5s, T per iter: 1.6s p/m  0.1s, T rest:  39.9s p/m  0.3s
|Iter: 131, T iter: 1.6s, T trans:  210.4s, T per iter: 1.6s p/m  0.1s, T rest:  31.9s p/m  0.3s
|Iter: 136, T iter: 1.6s, T trans:  218.3s, T per iter: 1.6s p/m  0.1s, T rest:  23.9s p/m  0.2s
|Iter: 141, T iter: 1.7s, T trans:  226.3s, T per iter: 1.6s p/m  0.1s, T rest:  15.9s p/m  0.2s
|Iter: 146, T iter: 1.6s, T trans:  234.2s, T per iter: 1.6s p/m  0.1s, T rest:  8.0s p/m  0.1s
</solve>
Wall time: 4min

%%time
segmentator.show_solution(
    show_iters=[45, 50, 65]
)
plt.tight_layout()

Wall time: 644 ms

path_save_file_per_test["test_4"] = segmentator.path_save_file

img_r = get_groud_truth(segmentator.id_name)
plt.imshow(img_r, cmap=ListedColormap(['black','tab:red']));

plot_precision_recall(img_r, segmentator.iterations)

%%time
id_name = pascal_images[2]
segmentator = LevelSetMethod(id_name=id_name)
segmentator.set_image(
    dic_img_arr[id_name], 
    reshape=True,
    norm=True
)
segmentator.show_image(grid=True)

<set_image> img: <class 'numpy.ndarray'>, reshape: True, norm: True
|img_dim:     [500 326]
|img_min/max: (0, 255)
|new_img_min/max(resize): (3.7037982399100635e-09, 1.1491087769469535e-07)
|new_dim:     [100  65]
|new_img_min/max(norm): (0, 255)
|size_domain: 6500
</set_image>
Wall time: 18 ms

%%time
segmentator.eval_stopfunct(
    list_sgm=np.array([0.01, 0.5, 1]), 
    list_lmbd=np.array([15, 10, 5]),
    cmap="viridis_r"
)
plt.tight_layout()

Wall time: 267 ms

%%time
segmentator.set_stopfunct(
    sgm=0.05, 
    lmbd=10,
)

<set_stopfunct> sigma: 0.05, lambda: 10
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (0.003928809963462067, 1.0)
</set_stopfunct>
Wall time: 1 ms

%%time
segmentator.eval_sgndist(
    c=np.array([0.4, 0.62]), 
    r=0.23,
    e=0.57
)

Wall time: 18 ms

%%time
segmentator.set_sgndist(
    c=np.array([0.4, 0.62]), 
    r=0.23,
    e=0.57
)

<set_sgndist> centro: [0.4  0.62], radio: 0.23, excentricidad: 0.57
</set_sgndist>
Wall time: 1 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2008_002247__iterations_times__sgm_0_05__lmbd_10__cx_0_4__cy_0_62__r_0_23__e_0_57__tau_0_1__k_-5
|Iter: 1, T iter: 1.4s, T trans:  2.4s, T per iter: 1.2s p/m  0.2s, T rest:  176.7s p/m  2.5s
|Iter: 6, T iter: 1.0s, T trans:  7.4s, T per iter: 1.1s p/m  0.1s, T rest:  154.0s p/m  1.6s
|Iter: 11, T iter: 1.1s, T trans:  12.7s, T per iter: 1.1s p/m  0.1s, T rest:  148.5s p/m  1.2s
|Iter: 16, T iter: 1.1s, T trans:  18.1s, T per iter: 1.1s p/m  0.1s, T rest:  143.7s p/m  1.0s
|Iter: 21, T iter: 1.1s, T trans:  23.4s, T per iter: 1.1s p/m  0.1s, T rest:  138.3s p/m  0.9s
|Iter: 26, T iter: 1.1s, T trans:  28.8s, T per iter: 1.1s p/m  0.1s, T rest:  133.3s p/m  0.8s
|Iter: 31, T iter: 1.1s, T trans:  34.2s, T per iter: 1.1s p/m  0.1s, T rest:  128.3s p/m  0.7s
|Iter: 36, T iter: 1.1s, T trans:  39.6s, T per iter: 1.1s p/m  0.1s, T rest:  123.2s p/m  0.7s
|Iter: 41, T iter: 1.1s, T trans:  45.2s, T per iter: 1.1s p/m  0.1s, T rest:  118.4s p/m  0.7s
|Iter: 46, T iter: 1.1s, T trans:  50.6s, T per iter: 1.1s p/m  0.1s, T rest:  113.1s p/m  0.6s
|Iter: 51, T iter: 1.1s, T trans:  56.0s, T per iter: 1.1s p/m  0.1s, T rest:  107.8s p/m  0.6s
|Iter: 56, T iter: 1.2s, T trans:  61.6s, T per iter: 1.1s p/m  0.1s, T rest:  102.6s p/m  0.6s
|Iter: 61, T iter: 1.1s, T trans:  67.2s, T per iter: 1.1s p/m  0.1s, T rest:  97.6s p/m  0.5s
|Iter: 66, T iter: 1.1s, T trans:  72.7s, T per iter: 1.1s p/m  0.1s, T rest:  92.3s p/m  0.5s
|Iter: 71, T iter: 1.1s, T trans:  78.3s, T per iter: 1.1s p/m  0.1s, T rest:  87.0s p/m  0.5s
|Iter: 76, T iter: 1.1s, T trans:  83.6s, T per iter: 1.1s p/m  0.1s, T rest:  81.4s p/m  0.5s
|Iter: 81, T iter: 1.0s, T trans:  88.8s, T per iter: 1.1s p/m  0.1s, T rest:  75.8s p/m  0.4s
|Iter: 86, T iter: 1.1s, T trans:  94.0s, T per iter: 1.1s p/m  0.1s, T rest:  70.2s p/m  0.4s
|Iter: 91, T iter: 1.1s, T trans:  99.4s, T per iter: 1.1s p/m  0.1s, T rest:  64.8s p/m  0.4s
|Iter: 96, T iter: 1.0s, T trans:  104.6s, T per iter: 1.1s p/m  0.1s, T rest:  59.3s p/m  0.4s
|Iter: 101, T iter: 1.1s, T trans:  109.9s, T per iter: 1.1s p/m  0.1s, T rest:  53.9s p/m  0.4s
|Iter: 106, T iter: 1.0s, T trans:  115.2s, T per iter: 1.1s p/m  0.1s, T rest:  48.5s p/m  0.3s
|Iter: 111, T iter: 1.1s, T trans:  120.6s, T per iter: 1.1s p/m  0.0s, T rest:  43.1s p/m  0.3s
|Iter: 116, T iter: 1.0s, T trans:  125.8s, T per iter: 1.1s p/m  0.0s, T rest:  37.6s p/m  0.3s
|Iter: 121, T iter: 1.1s, T trans:  131.0s, T per iter: 1.1s p/m  0.0s, T rest:  32.2s p/m  0.3s
|Iter: 126, T iter: 1.1s, T trans:  136.4s, T per iter: 1.1s p/m  0.0s, T rest:  26.8s p/m  0.2s
|Iter: 131, T iter: 1.1s, T trans:  141.9s, T per iter: 1.1s p/m  0.0s, T rest:  21.5s p/m  0.2s
|Iter: 136, T iter: 1.0s, T trans:  147.2s, T per iter: 1.1s p/m  0.0s, T rest:  16.1s p/m  0.2s
|Iter: 141, T iter: 1.2s, T trans:  152.7s, T per iter: 1.1s p/m  0.0s, T rest:  10.8s p/m  0.1s
|Iter: 146, T iter: 1.1s, T trans:  158.1s, T per iter: 1.1s p/m  0.0s, T rest:  5.4s p/m  0.1s
</solve>
Wall time: 2min 42s

%%time
segmentator.show_solution(
    show_iters=[45, 50, 150]
)
plt.tight_layout()

Wall time: 1.08 s

path_save_file_per_test["test_5"] = segmentator.path_save_file

img_r = get_groud_truth(segmentator.id_name)
plt.imshow(img_r, cmap=ListedColormap(['black','tab:red']));

plot_precision_recall(img_r, segmentator.iterations)

%%time
id_name = pascal_images[3]
segmentator = LevelSetMethod(id_name=id_name)
segmentator.set_image(
    dic_img_arr[id_name], 
    reshape=True,
    norm=True
)
segmentator.show_image(grid=True)

<set_image> img: <class 'numpy.ndarray'>, reshape: True, norm: True
|img_dim:     [319 500]
|img_min/max: (25, 255)
|new_img_min/max(resize): (2.584420145159685e-08, 1.1757947507258027e-07)
|new_dim:     [ 63 100]
|new_img_min/max(norm): (0, 255)
|size_domain: 6300
</set_image>
Wall time: 17 ms

%%time
segmentator.eval_stopfunct(
    list_sgm=np.array([0.01, 0.5, 1]), 
    list_lmbd=np.array([15, 5, 1]),
    cmap="viridis_r"
)
plt.tight_layout()

Wall time: 173 ms

segmentator.set_stopfunct(
    sgm=0.01, 
    lmbd=5
)

<set_stopfunct> sigma: 0.01, lambda: 5
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (0.0007370500309561012, 1.0)
</set_stopfunct>

%%time
segmentator.eval_sgndist(
    c=np.array([0.5, 0.5]), 
    r=0.14,
    e=1.8
)

Wall time: 18 ms

%%time
segmentator.set_sgndist(
    c=np.array([0.5, 0.5]), 
    r=0.14,
    e=1.8
)

<set_sgndist> centro: [0.5 0.5], radio: 0.14, excentricidad: 1.8
</set_sgndist>
Wall time: 1 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2007_004052__iterations_times__sgm_0_01__lmbd_5__cx_0_5__cy_0_5__r_0_14__e_1_8__tau_0_1__k_-5
|Iter: 1, T iter: 1.4s, T trans:  2.4s, T per iter: 1.2s p/m  0.2s, T rest:  176.5s p/m  2.7s
|Iter: 6, T iter: 1.0s, T trans:  7.3s, T per iter: 1.0s p/m  0.1s, T rest:  151.9s p/m  1.7s
|Iter: 11, T iter: 1.0s, T trans:  12.3s, T per iter: 1.0s p/m  0.1s, T rest:  143.8s p/m  1.3s
|Iter: 16, T iter: 1.0s, T trans:  17.4s, T per iter: 1.0s p/m  0.1s, T rest:  138.3s p/m  1.1s
|Iter: 21, T iter: 1.0s, T trans:  22.5s, T per iter: 1.0s p/m  0.1s, T rest:  132.8s p/m  1.0s
|Iter: 26, T iter: 1.0s, T trans:  27.5s, T per iter: 1.0s p/m  0.1s, T rest:  127.2s p/m  0.9s
|Iter: 31, T iter: 1.0s, T trans:  32.5s, T per iter: 1.0s p/m  0.1s, T rest:  121.9s p/m  0.8s
|Iter: 36, T iter: 1.0s, T trans:  37.4s, T per iter: 1.0s p/m  0.1s, T rest:  116.3s p/m  0.7s
|Iter: 41, T iter: 1.1s, T trans:  42.6s, T per iter: 1.0s p/m  0.1s, T rest:  111.4s p/m  0.7s
|Iter: 46, T iter: 1.0s, T trans:  47.5s, T per iter: 1.0s p/m  0.1s, T rest:  106.1s p/m  0.6s
|Iter: 51, T iter: 1.0s, T trans:  52.5s, T per iter: 1.0s p/m  0.1s, T rest:  101.0s p/m  0.6s
|Iter: 56, T iter: 1.0s, T trans:  57.5s, T per iter: 1.0s p/m  0.1s, T rest:  95.9s p/m  0.5s
|Iter: 61, T iter: 1.0s, T trans:  62.5s, T per iter: 1.0s p/m  0.1s, T rest:  90.7s p/m  0.5s
|Iter: 66, T iter: 1.0s, T trans:  67.4s, T per iter: 1.0s p/m  0.1s, T rest:  85.5s p/m  0.5s
|Iter: 71, T iter: 1.0s, T trans:  72.4s, T per iter: 1.0s p/m  0.1s, T rest:  80.4s p/m  0.5s
|Iter: 76, T iter: 1.0s, T trans:  77.3s, T per iter: 1.0s p/m  0.0s, T rest:  75.3s p/m  0.4s
|Iter: 81, T iter: 1.0s, T trans:  82.3s, T per iter: 1.0s p/m  0.0s, T rest:  70.3s p/m  0.4s
|Iter: 86, T iter: 1.0s, T trans:  87.3s, T per iter: 1.0s p/m  0.0s, T rest:  65.2s p/m  0.4s
|Iter: 91, T iter: 1.0s, T trans:  92.2s, T per iter: 1.0s p/m  0.0s, T rest:  60.2s p/m  0.4s
|Iter: 96, T iter: 1.0s, T trans:  97.4s, T per iter: 1.0s p/m  0.0s, T rest:  55.2s p/m  0.3s
|Iter: 101, T iter: 1.0s, T trans:  102.5s, T per iter: 1.0s p/m  0.0s, T rest:  50.3s p/m  0.3s
|Iter: 106, T iter: 1.1s, T trans:  107.7s, T per iter: 1.0s p/m  0.0s, T rest:  45.3s p/m  0.3s
|Iter: 111, T iter: 1.1s, T trans:  112.8s, T per iter: 1.0s p/m  0.0s, T rest:  40.3s p/m  0.3s
|Iter: 116, T iter: 1.0s, T trans:  117.8s, T per iter: 1.0s p/m  0.0s, T rest:  35.2s p/m  0.3s
|Iter: 121, T iter: 1.1s, T trans:  122.8s, T per iter: 1.0s p/m  0.0s, T rest:  30.2s p/m  0.2s
|Iter: 126, T iter: 1.0s, T trans:  127.9s, T per iter: 1.0s p/m  0.0s, T rest:  25.2s p/m  0.2s
|Iter: 131, T iter: 1.0s, T trans:  132.9s, T per iter: 1.0s p/m  0.0s, T rest:  20.1s p/m  0.2s
|Iter: 136, T iter: 1.0s, T trans:  137.8s, T per iter: 1.0s p/m  0.0s, T rest:  15.1s p/m  0.2s
|Iter: 141, T iter: 1.1s, T trans:  142.9s, T per iter: 1.0s p/m  0.0s, T rest:  10.1s p/m  0.1s
|Iter: 146, T iter: 1.0s, T trans:  147.8s, T per iter: 1.0s p/m  0.0s, T rest:  5.0s p/m  0.1s
</solve>
Wall time: 2min 31s

%%time
segmentator.show_solution(
    show_iters=[45, 50, 150]
)
plt.tight_layout()

Wall time: 86.1 ms

path_save_file_per_test["test_6"] = segmentator.path_save_file

img_r = get_groud_truth(segmentator.id_name)
plt.imshow(img_r, cmap=ListedColormap(['black','tab:red']));

plot_precision_recall(img_r, segmentator.iterations)

%%time
id_name = pascal_images[4]
segmentator = LevelSetMethod(id_name=id_name)
segmentator.set_image(
    dic_img_arr[id_name], 
    reshape=True,
    norm=True
)
segmentator.show_image(grid=True)

<set_image> img: <class 'numpy.ndarray'>, reshape: True, norm: True
|img_dim:     [358 500]
|img_min/max: (0, 255)
|new_img_min/max(resize): (2.561137080788877e-09, 1.1026334597240103e-07)
|new_dim:     [ 71 100]
|new_img_min/max(norm): (0, 255)
|size_domain: 7100
</set_image>
Wall time: 17.5 ms

%%time
segmentator.eval_stopfunct(
    list_sgm=np.array([0.01, 0.5, 1]), 
    list_lmbd=np.array([20, 10, 5]),
    cmap="viridis_r"
)
plt.tight_layout()

Wall time: 234 ms

segmentator.set_stopfunct(
    sgm=0.01, 
    lmbd=10
)

<set_stopfunct> sigma: 0.01, lambda: 10
|gauss_min/max: (0, 255)
|stopfunct_min/max:     (0.00261882938326568, 1.0)
</set_stopfunct>

%%time
segmentator.eval_sgndist(
    c=np.array([0.5, 0.54]), 
    r=0.14,
    e=1.5
)

Wall time: 20 ms

%%time
segmentator.set_sgndist(
    c=np.array([0.5, 0.54]), 
    r=0.14,
    e=1.5
)

<set_sgndist> centro: [0.5  0.54], radio: 0.14, excentricidad: 1.5
</set_sgndist>
Wall time: 1 ms

%%time
segmentator.solve(
    n_iter = 150,
    tau = 0.1,
    k = -5,
    save_rate=10,
    root_path_save="results",
    show_log_rate=5
)

<solve> N° iteraciones: 150, discretizacion en tiempo: 0.1, constante k: -5
|path_save_file: results/2009_005078__iterations_times__sgm_0_01__lmbd_10__cx_0_5__cy_0_54__r_0_14__e_1_5__tau_0_1__k_-5
|Iter: 1, T iter: 1.7s, T trans:  3.2s, T per iter: 1.6s p/m  0.1s, T rest:  236.9s p/m  1.8s
|Iter: 6, T iter: 1.3s, T trans:  9.5s, T per iter: 1.4s p/m  0.2s, T rest:  195.9s p/m  2.0s
|Iter: 11, T iter: 1.2s, T trans:  15.8s, T per iter: 1.3s p/m  0.1s, T rest:  184.2s p/m  1.6s
|Iter: 16, T iter: 1.3s, T trans:  22.2s, T per iter: 1.3s p/m  0.1s, T rest:  176.1s p/m  1.3s
|Iter: 21, T iter: 1.3s, T trans:  28.7s, T per iter: 1.3s p/m  0.1s, T rest:  169.6s p/m  1.1s
|Iter: 26, T iter: 1.3s, T trans:  35.3s, T per iter: 1.3s p/m  0.1s, T rest:  163.5s p/m  1.0s
|Iter: 31, T iter: 1.3s, T trans:  41.8s, T per iter: 1.3s p/m  0.1s, T rest:  156.7s p/m  0.9s
|Iter: 36, T iter: 1.3s, T trans:  48.3s, T per iter: 1.3s p/m  0.1s, T rest:  150.3s p/m  0.8s
|Iter: 41, T iter: 1.3s, T trans:  54.8s, T per iter: 1.3s p/m  0.1s, T rest:  143.6s p/m  0.8s
|Iter: 46, T iter: 1.3s, T trans:  61.1s, T per iter: 1.3s p/m  0.1s, T rest:  136.6s p/m  0.7s
|Iter: 51, T iter: 1.3s, T trans:  67.5s, T per iter: 1.3s p/m  0.1s, T rest:  129.9s p/m  0.7s
|Iter: 56, T iter: 1.3s, T trans:  73.9s, T per iter: 1.3s p/m  0.1s, T rest:  123.1s p/m  0.6s
|Iter: 61, T iter: 1.3s, T trans:  80.2s, T per iter: 1.3s p/m  0.1s, T rest:  116.4s p/m  0.6s
|Iter: 66, T iter: 1.2s, T trans:  86.5s, T per iter: 1.3s p/m  0.1s, T rest:  109.7s p/m  0.6s
|Iter: 71, T iter: 1.3s, T trans:  92.8s, T per iter: 1.3s p/m  0.1s, T rest:  103.1s p/m  0.6s
|Iter: 76, T iter: 1.2s, T trans:  99.0s, T per iter: 1.3s p/m  0.1s, T rest:  96.4s p/m  0.5s
|Iter: 81, T iter: 1.3s, T trans:  105.3s, T per iter: 1.3s p/m  0.1s, T rest:  89.9s p/m  0.5s
|Iter: 86, T iter: 1.2s, T trans:  111.5s, T per iter: 1.3s p/m  0.1s, T rest:  83.3s p/m  0.5s
|Iter: 91, T iter: 1.3s, T trans:  117.5s, T per iter: 1.3s p/m  0.1s, T rest:  76.7s p/m  0.5s
|Iter: 96, T iter: 1.2s, T trans:  123.5s, T per iter: 1.3s p/m  0.1s, T rest:  70.1s p/m  0.4s
|Iter: 101, T iter: 1.4s, T trans:  129.7s, T per iter: 1.3s p/m  0.1s, T rest:  63.6s p/m  0.4s
|Iter: 106, T iter: 1.2s, T trans:  136.0s, T per iter: 1.3s p/m  0.1s, T rest:  57.2s p/m  0.4s
|Iter: 111, T iter: 1.3s, T trans:  142.1s, T per iter: 1.3s p/m  0.1s, T rest:  50.7s p/m  0.4s
|Iter: 116, T iter: 1.3s, T trans:  148.3s, T per iter: 1.3s p/m  0.1s, T rest:  44.4s p/m  0.3s
|Iter: 121, T iter: 1.3s, T trans:  154.8s, T per iter: 1.3s p/m  0.1s, T rest:  38.1s p/m  0.3s
|Iter: 126, T iter: 1.2s, T trans:  160.9s, T per iter: 1.3s p/m  0.1s, T rest:  31.7s p/m  0.3s
|Iter: 131, T iter: 1.3s, T trans:  167.1s, T per iter: 1.3s p/m  0.1s, T rest:  25.3s p/m  0.3s
|Iter: 136, T iter: 1.2s, T trans:  173.2s, T per iter: 1.3s p/m  0.1s, T rest:  19.0s p/m  0.2s
|Iter: 141, T iter: 1.4s, T trans:  179.7s, T per iter: 1.3s p/m  0.1s, T rest:  12.7s p/m  0.2s
|Iter: 146, T iter: 1.3s, T trans:  186.1s, T per iter: 1.3s p/m  0.1s, T rest:  6.3s p/m  0.1s
</solve>
Wall time: 3min 11s

%%time
segmentator.show_solution(
    show_iters=[35, 50, 75]
)
plt.tight_layout()

Wall time: 1.18 s

path_save_file_per_test["test_7"] = segmentator.path_save_file

img_r = get_groud_truth(segmentator.id_name)
plt.imshow(img_r, cmap=ListedColormap(['black','tab:red']));

plot_precision_recall(img_r, segmentator.iterations)

Graficos

Estaticos

%%time
test_k = "test_5"
path_save_file = path_save_file_per_test[test_k]
id_name = path_save_file.split("__")[0].split("/")[1]
print(path_save_file)
iterations, times = pickle.load(open(path_save_file_per_test[test_k]+".pickle", "rb"))

img_r = get_groud_truth(id_name)
img_o = Image.open(f"dataset/database_pascal/{id_name}.jpg")

fig, ax = plt.subplots(2, 2, figsize=(2*3.5, 3.5*2), constrained_layout=False)
        
cmap_sgndist = "RdGy_r"
cmap_pr = "winter_r"

clf = [(un>0).astype(int) for un in iterations]

precision_recall_curve = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf]
precision = np.array([prc_k[0] for prc_k in precision_recall_curve])
recall = np.array([prc_k[1] for prc_k in precision_recall_curve])

mask_0 = np.isclose(precision+recall, 0)
mask_1 = ~mask_0
    
best_iter= np.argmax( mask_1*(2*precision*recall)/(precision+recall+mask_0) )

ax[0,0].imshow(img_o)
    
ax[0,1].imshow(img_r, cmap=ListedColormap(['black','tab:red']))

ax[1,1].imshow(clf[best_iter], cmap=ListedColormap(['black','tab:red']))

ax[1,0].grid()
ax[1,0].scatter(precision, recall, c=range(len(precision_recall_curve)), cmap=cmap_pr)
ax[1,0].scatter([precision[best_iter]], [recall[best_iter]], c="red", s=50, label=f"Iter: {best_iter}\nRecall: {100*recall[best_iter]: .1f}\%\nPrecision: {100*precision[best_iter]: .1f}\%")

ax[1,0].set_yticks(np.linspace(3/4, 1, 5).round(2));
ax[1,0].set_xticks(np.linspace(3/4, 1, 5).round(2));
    
plt_eps = 0.01
plt.xticks(fontsize=18*0.8)
plt.yticks(fontsize=18*0.8)
    
ax[1,0].set_xlim([3/4-plt_eps, 1+plt_eps])
ax[1,0].set_ylim([3/4-plt_eps, 1+plt_eps])
    

ax[1,0].legend(fontsize=18*0.8)
    
ax[1,0].set_aspect('equal')
    

ax[0,0].set_xticks([])
ax[0,0].set_yticks([])

ax[0,1].set_xticks([])
ax[0,1].set_yticks([])

ax[1,1].set_xticks([])
ax[1,1].set_yticks([]);


ax[1,1].set_title("Conjunto de nivel\npositivo", size=18)
ax[1,1].set_xlabel(f"Iteración {best_iter}", size=18)

ax[0,1].set_title("Target", size=18)
ax[0,0].set_title("Imagen original", size=18)
# ax[1,1].set_xlabel(f"Iteración {best_iter}", size=18)

ax[1,0].set_ylabel("Recall", fontsize=18)
ax[1,0].set_xlabel("Precision", fontsize=18)
ax[1,0].set_title("Precision vs Recall", size=18)

plt.tight_layout()
fig.savefig(f"PASCAL_{test_k}.pdf", bbox_inches='tight')

results/2008_002247__iterations_times__sgm_0_05__lmbd_10__cx_0_4__cy_0_62__r_0_23__e_0_57__tau_0_1__k_-5
Wall time: 348 ms

Dinamicos

%%time
iterations_0, _ = pickle.load(open(path_save_file_per_test["test_0"]+".pickle", "rb"))
iterations_1, _ = pickle.load(open(path_save_file_per_test["test_1"]+".pickle", "rb"))
iterations_2, _ = pickle.load(open(path_save_file_per_test["test_2"]+".pickle", "rb"))
iterations_3, _ = pickle.load(open(path_save_file_per_test["test_3"]+".pickle", "rb"))
print(path_save_file_per_test["test_0"], path_save_file_per_test["test_1"], path_save_file_per_test["test_2"], path_save_file_per_test["test_3"], sep="\n")

results/2009_001314__iterations_times__sgm_0_01__lmbd_5__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
results/2009_001314__iterations_times__sgm_0_01__lmbd_15__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
results/2009_001314__iterations_times__sgm_0_01__lmbd_1__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
results/2009_001314__iterations_times__sgm_1__lmbd_5__cx_0_4__cy_0_5__r_0_18__e_1__tau_0_1__k_-5
Wall time: 24 ms

id_name = pascal_images[0]
dim = [91, 100]
cmap_sgndist = "RdGy_r"
img_r = get_groud_truth(id_name)
img_o = Image.open(f"dataset/database_pascal/{id_name}.jpg")

clf_0 = [(un>0).astype(int) for un in iterations_0]
clf_1 = [(un>0).astype(int) for un in iterations_1]
clf_2 = [(un>0).astype(int) for un in iterations_2]
clf_3 = [(un>0).astype(int) for un in iterations_3]

precision_recall_curve_0 = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf_0]
precision_0 = np.array([prc_k[0] for prc_k in precision_recall_curve_0])
recall_0 = np.array([prc_k[1] for prc_k in precision_recall_curve_0])

precision_recall_curve_1 = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf_1]
precision_1 = np.array([prc_k[0] for prc_k in precision_recall_curve_1])
recall_1 = np.array([prc_k[1] for prc_k in precision_recall_curve_1])

precision_recall_curve_2 = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf_2]
precision_2 = np.array([prc_k[0] for prc_k in precision_recall_curve_2])
recall_2 = np.array([prc_k[1] for prc_k in precision_recall_curve_2])

precision_recall_curve_3 = [precision_recall(img_r, clf_k, pos_label=1) for clf_k in clf_3]
precision_3 = np.array([prc_k[0] for prc_k in precision_recall_curve_3])
recall_3 = np.array([prc_k[1] for prc_k in precision_recall_curve_3])

# %%time
fig = plt.figure(figsize=(8*3,8*2))

cmap_pr = "hot_r"
plt_eps = 0.01

iter_k = 55


sgm_lmbd_0 = (0.01, 5)  #0
sgm_lmbd_1 = (0.01, 15) #1

zz_0 = iterations_0[iter_k]
zz_1 = iterations_1[iter_k]

p_0 = precision_0
r_0 = recall_0
p_1 = precision_1
r_1 = recall_1
 
    
# sgm_lmbd_0 = (0.01, 5)  #0
# sgm_lmbd_1 = (0.01, 1)  #2

# zz_0 = iterations_0[iter_k]
# zz_1 = iterations_2[iter_k]

# p_0 = precision_0
# r_0 = recall_0
# p_1 = precision_2
# r_1 = recall_2


# sgm_lmbd_0 = (0.01, 5)  #0
# sgm_lmbd_1 = (1, 5)     #3

# zz_0 = iterations_0[iter_k]
# zz_1 = iterations_3[iter_k]

# p_0 = precision_0
# r_0 = recall_0
# p_1 = precision_3
# r_1 = recall_3


x = np.linspace(0, 1, dim[1])
y = np.linspace(0, 1, dim[0])
xx, yy = np.meshgrid(x, y)

ax1 = fig.add_subplot(231)
ax2 = fig.add_subplot(232)
ax3 = fig.add_subplot(233)

ax4 = fig.add_subplot(234)
ax5 = fig.add_subplot(235)
ax6 = fig.add_subplot(236)

ax1.imshow(img_o)
    
ax4.imshow(img_r, cmap=ListedColormap(['black','tab:red']))

ax2.imshow((zz_0>0), cmap=ListedColormap(['black','tab:red']))

ax5.imshow((zz_1>0), cmap=ListedColormap(['black','tab:red']))

ax3.scatter(p_0[:iter_k], r_0[:iter_k], c=range(iter_k), cmap=cmap_pr, s=50*3)

ax6.scatter(p_1[:iter_k], r_0[:iter_k], c=range(iter_k), cmap=cmap_pr, s=50*3)

ax1.set_xticks([])
ax1.set_yticks([])

ax4.set_xticks([])
ax4.set_yticks([])

ax2.set_xticks([])
ax2.set_yticks([])

ax5.set_xticks([])
ax5.set_yticks([])

ax3.set_yticks(np.linspace(3/4, 1, 5).round(2));
ax3.set_xticks(np.linspace(3/4, 1, 5).round(2));
ax3.set_xticklabels(["" for _ in range(5)]);
ax3.tick_params(labelsize = 24)
ax3.grid()
ax3.set_xlim([3/4-plt_eps, 1+plt_eps])
ax3.set_ylim([3/4-plt_eps, 1+plt_eps])
ax3.set_ylabel("Recall", fontsize=18*2, rotation=270)
ax3.yaxis.set_label_coords(1.01, 0.5)

ax6.set_yticks(np.linspace(3/4, 1, 5).round(2));
ax6.set_xticks(np.linspace(3/4, 1, 5).round(2));
ax6.tick_params(labelsize = 24)
ax6.grid()
ax6.set_xlim([3/4-plt_eps, 1+plt_eps])
ax6.set_ylim([3/4-plt_eps, 1+plt_eps])
ax6.set_ylabel("Recall", fontsize=18*2, rotation=270)
ax6.yaxis.set_label_coords(1.01, 0.5)
ax6.set_xlabel("Precision", fontsize=18*2)

ax1.set_title(id_name, size=18*2)
ax2.set_title("Conjunto de nivel\npositivo", size=18*2)
ax3.set_title("Precision vs Recall", size=18*2)

ax1.set_ylabel("Imagen original", size=18*2)
ax4.set_ylabel("Target", size=18*2)

ax2.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_0)}", size=18*2)
ax5.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_1)}", size=18*2)

ax5.set_xlabel(f"Iteración {iter_k}", size=18*3)
ax5.xaxis.set_label_coords(0.5, -0.1)

plt.tight_layout()

# %%time
fps = 250
frn = 150

id_test = "0_3"

# sgm_lmbd_0 = (0.01, 5)  #0
# sgm_lmbd_1 = (0.01, 15) #1

# zz_0 = iterations_0
# zz_1 = iterations_1

# p_0 = precision_0
# r_0 = recall_0
# p_1 = precision_1
# r_1 = recall_1
 
    
# sgm_lmbd_0 = (0.01, 5)  #0
# sgm_lmbd_1 = (0.01, 1)  #2

# zz_0 = iterations_0
# zz_1 = iterations_2

# p_0 = precision_0
# r_0 = recall_0
# p_1 = precision_2
# r_1 = recall_2


sgm_lmbd_0 = (0.01, 5)  #0
sgm_lmbd_1 = (1, 5)     #3

zz_0 = iterations_0
zz_1 = iterations_3

p_0 = precision_0
r_0 = recall_0
p_1 = precision_3
r_1 = recall_3


zarray_0 = np.zeros((dim[0], dim[1], frn))
zarray_1 = np.zeros((dim[0], dim[1], frn))

parray_0 = np.zeros(frn+1)
parray_1 = np.zeros(frn+1)

rarray_0 = np.zeros(frn+1)
rarray_1 = np.zeros(frn+1)

for i in range(frn):
    zarray_0[:, :, i] = zz_0[i]
    zarray_1[:, :, i] = zz_1[i]
    
    parray_0[i+1] = p_0[i]
    parray_1[i+1] = p_1[i]

    rarray_0[i+1] = r_0[i]
    rarray_1[i+1] = r_1[i]

zarray = np.array([zarray_0, zarray_1])
parray = np.array([parray_0, parray_1])
rarray = np.array([rarray_0, rarray_1])

def change_plot(frame_number, zarray, parray, rarray, plot):
    plot[0].remove()
    plot[0] = ax1.imshow(img_o)

    plot[1].remove()
    plot[1] = ax2.imshow((zarray[0][:, :, frame_number]>0), cmap=ListedColormap(['black','tab:red']))

    plot[2].remove()
    plot[2] = ax3.scatter(parray[0][:frame_number+1], rarray[0][:frame_number+1], c=range(frame_number+1), cmap=cmap_pr, s=50*3)

    plot[3].remove()
    plot[3] = ax4.imshow(img_r, cmap=ListedColormap(['black','tab:red']))

    plot[4].remove()
    plot[4] = ax5.imshow((zarray[1][:, :, frame_number]>0), cmap=ListedColormap(['black','tab:red']))

    plot[5].remove()
    plot[5] = ax6.scatter(parray[1][:frame_number+1], rarray[1][:frame_number+1], c=range(frame_number+1), cmap=cmap_pr, s=50*3)
    
    ax1.set_xticks([])
    ax1.set_yticks([])

    ax4.set_xticks([])
    ax4.set_yticks([])

    ax2.set_xticks([])
    ax2.set_yticks([])

    ax5.set_xticks([])
    ax5.set_yticks([])

    ax3.set_yticks(np.linspace(3/4, 1, 5).round(2));
    ax3.set_xticks(np.linspace(3/4, 1, 5).round(2));
    ax3.set_xticklabels(["" for _ in range(5)]);
    ax3.tick_params(labelsize = 24)
#     ax3.grid()
    ax3.set_xlim([3/4-plt_eps, 1+plt_eps])
    ax3.set_ylim([3/4-plt_eps, 1+plt_eps])
    ax3.set_ylabel("Recall", fontsize=18*2, rotation=270)
    ax3.yaxis.set_label_coords(1.01, 0.5)

    ax6.set_yticks(np.linspace(3/4, 1, 5).round(2));
    ax6.set_xticks(np.linspace(3/4, 1, 5).round(2));
    ax6.tick_params(labelsize = 24)
#     ax6.grid()
    ax6.set_xlim([3/4-plt_eps, 1+plt_eps])
    ax6.set_ylim([3/4-plt_eps, 1+plt_eps])
    ax6.set_ylabel("Recall", fontsize=18*2, rotation=270)
    ax6.yaxis.set_label_coords(1.01, 0.5)
    ax6.set_xlabel("Precision", fontsize=18*2)

    ax1.set_title(id_name, size=18*2)
    ax2.set_title("Conjunto de nivel\npositivo", size=18*2)
    ax3.set_title("Precision vs Recall", size=18*2)

    ax1.set_ylabel("Imagen original", size=18*2)
    ax4.set_ylabel("Target", size=18*2)

    ax2.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_0)}", size=18*2)
    ax5.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_1)}", size=18*2)

    ax5.set_xlabel(f"Iteración {frame_number}", size=18*3)
    ax5.xaxis.set_label_coords(0.5, -0.1)

    
fig = plt.figure(figsize=(8*3,8*2))

ax1 = fig.add_subplot(231)
ax2 = fig.add_subplot(232)
ax3 = fig.add_subplot(233)

ax4 = fig.add_subplot(234)
ax5 = fig.add_subplot(235)
ax6 = fig.add_subplot(236)

plot = [
    ax1.imshow(img_o),
    ax4.imshow(img_r, cmap=ListedColormap(['black','tab:red'])),
    ax2.imshow((zarray[0][:, :, 0]>0), cmap=ListedColormap(['black','tab:red'])),
    ax5.imshow((zarray[1][:, :, 0]>0), cmap=ListedColormap(['black','tab:red'])),
    ax3.scatter(parray[0][:1], rarray[0][:1], c=range(1), cmap=cmap_pr, s=50*3),
    ax6.scatter(parray[1][:1], rarray[1][:1], c=range(1), cmap=cmap_pr, s=50*3)
]

ax1.set_xticks([])
ax1.set_yticks([])

ax4.set_xticks([])
ax4.set_yticks([])

ax2.set_xticks([])
ax2.set_yticks([])

ax5.set_xticks([])
ax5.set_yticks([])

ax3.set_yticks(np.linspace(3/4, 1, 5).round(2));
ax3.set_xticks(np.linspace(3/4, 1, 5).round(2));
ax3.set_xticklabels(["" for _ in range(5)]);
ax3.tick_params(labelsize = 24)
# ax3.grid()
ax3.set_xlim([3/4-plt_eps, 1+plt_eps])
ax3.set_ylim([3/4-plt_eps, 1+plt_eps])
ax3.set_ylabel("Recall", fontsize=18*2, rotation=270)
ax3.yaxis.set_label_coords(1.01, 0.5)

ax6.set_yticks(np.linspace(3/4, 1, 5).round(2));
ax6.set_xticks(np.linspace(3/4, 1, 5).round(2));
ax6.tick_params(labelsize = 24)
# ax6.grid()
ax6.set_xlim([3/4-plt_eps, 1+plt_eps])
ax6.set_ylim([3/4-plt_eps, 1+plt_eps])
ax6.set_ylabel("Recall", fontsize=18*2, rotation=270)
ax6.yaxis.set_label_coords(1.01, 0.5)
ax6.set_xlabel("Precision", fontsize=18*2)

ax1.set_title(id_name, size=18*2)
ax2.set_title("Conjunto de nivel\npositivo", size=18*2)
ax3.set_title("Precision vs Recall", size=18*2)

ax1.set_ylabel("Imagen original", size=18*2)
ax4.set_ylabel("Target", size=18*2)

ax2.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_0)}", size=18*2)
ax5.set_ylabel(f"(sigma, lambda) = {str(sgm_lmbd_1)}", size=18*2)

ax5.set_xlabel(f"Iteración {iter_k}", size=18*3)
ax5.xaxis.set_label_coords(0.5, -0.1)

ani = animation.FuncAnimation(fig, change_plot, frn, fargs=(zarray, parray, rarray, plot), interval=1000 / fps)

path_name_gif = f"test_pascal_{id_test}"
ani.save(f'{path_name_gif}.gif');

MovieWriter ffmpeg unavailable; using Pillow instead.

mygif = Image.open(path_name_gif+".gif")

for frame in range(0, mygif.n_frames):
    mygif.seek(frame)
    mygif.save(f"{path_name_gif}_gif/"+"frame-"+str(frame)+".png")